South KoreaManufacturing2025 – ongoingShinhan Diamond
AI inspection of CMP disks for wafer polishing
A maker of chip-polishing consumables cut quality inspection time by 33% with automated AI inspection
Problem
Shinhan Diamond makes CMP disks, consumables used in the chemical-mechanical polishing of semiconductor wafers. Quality inspection of the disks is a time-consuming step before shipment.
Approach
Within the Ministry of Trade, Industry and Energy's AI Factory leading project, the company introduced an AI-based automatic inspection system for CMP disks. The ministry included the result in its August 2026 review of the 42 earliest of about 170 supported sites.
Results
Takeaway
Small manufacturers faced greater difficulty keeping AI running after support ended than deploying it initially, according to later press analysis. The ministry gave no baseline or measurement method for Shinhan Diamond's inspection-time reduction.
Categories
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Government support program
- Evidence
- Company's own claim
Sources
Source summaries
The ministry reviewed the AI Factory leading project. After reviewing 42 of about 170 supported sites, it reports average productivity up 30.1% and defect rates down 15.5%, and lists company examples, including Shinhan Diamond's 33% cut in quality inspection time for wafer-polishing consumables. The ministry gave no measurement details.
Korea Data Economy News analysed vision inspection and predictive maintenance results in Korean factories. It specifies that Shinhan Diamond's system inspects CMP disks and repeats the 33% figure. It cautions that the reviewed sites were early performers and not an industry average.
Surveys AI results at small and mid-sized manufacturers, citing the ministry's figures for several firms, including Shinhan Diamond's 33% shorter inspection time. It stresses that results vary by company and process and that staffing, operating burden and maintaining systems after deployment remain the main difficulties.
Full details are in the original sources.